Re: Ada advocacy

["Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de>]

On Wed, 28 Aug 2013 18:44:38 +0200, Yannick Duchêne (Hibou57) wrote:

> Le Wed, 28 Aug 2013 13:42:16 +0200, G.B.  
> <rm-dash-bau-haus@dash.futureapps.de> a écrit:
>>
>> That's the deal, a reference to an order of events in time,
>> right in the type. Something that ordinary types don't have.
> 
> As a comment aside, that's why it happened I missed Ada had coroutines  
> (using a task in place of a coroutine is too much).

It is an interesting question, but too vague.

1. preemptive or not?

2. implicit state or explicit (bound to some object)

3. how much of the context where given execution path is nested would be
accessible?

4. inheritance issues

Maybe instead of coroutines... Ada does not specify whether tasks should be
native. What about providing tasks scheduled by the RTL (or user), in
addition to the native tasks, not as a replacement. Such tasks would have
many limitations regarding I/O, but no more coroutines would. And they will
be lighter than native tasks.

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de